if key is in index, use it for distinct SERVER-1673

db/clientcursor.cpp

@@ -258,8 +258,7 @@ namespace mongo {
         _cursorid = allocCursorId_inlock();
         clientCursorsById.insert( make_pair(_cursorid, this) );
         
-#if 0  
-        { 
+        if ( ! _c->modifiedKeys() ) { 
             // store index information so we can decide if we can 
             // get something out of the index key rather than full object
             
@@ -274,7 +273,7 @@ namespace mongo {
                 x++;
             }
         }
-#endif
+
     }
     
 
@@ -292,6 +291,24 @@ namespace mongo {
         }
     }
 
+    bool ClientCursor::getFieldsDotted( const string& name, BSONElementSet &ret ) {
+        
+        map<string,int>::const_iterator i = _indexedFields.find( name );
+        if ( i == _indexedFields.end() ){
+            current().getFieldsDotted( name , ret );
+            return false;
+        }
+
+        int x = i->second;
+        
+        BSONObjIterator it( currKey() );
+        while ( x && it.more() )
+            it.next();
+        assert( x == 0 );
+        ret.insert( it.next() );
+        return true;
+    }
+
     /* call when cursor's location changes so that we can update the
        cursorsbylocation map.  if you are locked and internally iterating, only
        need to call when you are ready to "unlock".

db/clientcursor.h

@@ -221,7 +221,15 @@ namespace mongo {
         bool advance(){ return _c->advance(); }
         BSONObj current() { return _c->current(); }
         DiskLoc currLoc() { return _c->currLoc(); }
-        BSONObj currKey() { return _c->currKey(); }
+        BSONObj currKey() const { return _c->currKey(); }
+        
+
+        /**
+         * same as BSONObj::getFieldsDotted
+         * if it can be retrieved from key, it is
+         * @return if this was retrieved from key
+         */
+        bool getFieldsDotted( const string& name, BSONElementSet &ret );
 
         bool currentIsDup() { return _c->getsetdup( _c->currLoc() ); }
         

db/commands/distinct.cpp

@@ -47,8 +47,9 @@ namespace mongo {
             BSONArrayBuilder arr( bb );
             BSONElementSet values;
             
-            long long nscanned = 0;
-            long long n = 0;
+            long long nscanned = 0; // locations looked at
+            long long nscannedObjects = 0; // full objects looked at
+            long long n = 0; // matches
             MatchDetails md;
 
             shared_ptr<Cursor> cursor = bestGuessCursor(ns.c_str() , query , BSONObj() );
@@ -59,11 +60,10 @@ namespace mongo {
                 bool loadedObject = false;
                 
                 if ( !cursor->matcher() || cursor->matcher()->matchesCurrent( cursor.get() , &md ) ){
-                    loadedObject = true;
-                    BSONObj o = cursor->current();
+                    n++;
 
                     BSONElementSet temp;
-                    o.getFieldsDotted( key, temp );
+                    loadedObject = ! cc->getFieldsDotted( key , temp );
                     
                     for ( BSONElementSet::iterator i=temp.begin(); i!=temp.end(); ++i ){
                         BSONElement e = *i;
@@ -82,7 +82,7 @@ namespace mongo {
                 }
 
                 if ( loadedObject || md.loadedObject ) 
-                    n++;
+                    nscannedObjects++;
 
                 cursor->advance();
 
@@ -100,6 +100,7 @@ namespace mongo {
                 BSONObjBuilder b;
                 b.appendNumber( "n" , n );
                 b.appendNumber( "nscanned" , nscanned );
+                b.appendNumber( "nscannedObjects" , nscannedObjects );
                 result.append( "stats" , b.obj() );
             }
             

jstests/distinct_index1.js

@@ -2,8 +2,8 @@
 t = db.distinct_index1
 t.drop();
 
-function r(){
-    return Math.floor( Math.random() * 10 );
+function r( x ){
+    return Math.floor( Math.sqrt( x * 123123 ) ) % 10;
 }
 
 function d( k , q ){
@@ -11,17 +11,32 @@ function d( k , q ){
 }
 
 for ( i=0; i<1000; i++ ){
-    o = { a : r() , b : r() };
+    o = { a : r(i*5) , b : r(i) };
     t.insert( o );
 }
 
 x = d( "a" );
 assert.eq( 1000 , x.stats.n , "A1" )
 assert.eq( 1000 , x.stats.nscanned , "A2" )
+assert.eq( 1000 , x.stats.nscannedObjects , "A3" )
 
 x = d( "a" , { a : { $gt : 5 } } );
-assert.eq( 1000 , x.stats.n , "A1" )
-assert.eq( 1000 , x.stats.nscanned , "A2" )
+assert.eq( 398 , x.stats.n , "B1" )
+assert.eq( 1000 , x.stats.nscanned , "B2" )
+assert.eq( 1000 , x.stats.nscannedObjects , "B3" )
+
+t.ensureIndex( { a : 1 } )
+
+x = d( "a" );
+assert.eq( 1000 , x.stats.n , "C1" )
+assert.eq( 1000 , x.stats.nscanned , "C2" )
+assert.eq( 1000 , x.stats.nscannedObjects , "C3" )
+
+x = d( "a" , { a : { $gt : 5 } } );
+assert.eq( 398 , x.stats.n , "D1" )
+assert.eq( 398 , x.stats.nscanned , "D2" )
+assert.eq( 0 , x.stats.nscannedObjects , "D3" )
+